Passive islanding detection method using Wavelet Packet Transform in Grid Connected Photovoltaic Systems

In this paper, a new approach in the passive-islanding detection method has been studied for the grid-connected photovoltaic (PV) inverter system using the wavelet packet transform (WPT) and back propagation neural network (BPNN). This proposed method is based on the point of common coupling voltage measurement and processing of this signal with a WPT to find the normalized logarithmic energy entropy. The BPNN is used for decision-making mechanism to avoid threshold. With this technique, the islanding operation is able to detect within 40 ms. In addition, it is effective even when the power supplied by distributed generation and the power consumed by the load is matched. In order to verify the capability of the proposed method, the simulation and experimental results of a 1.2-kW PV inverter by IEC Std. 62116 are provided. The proposed technique can be applied to both single- and multi-PV systems.